1/1/2023

Fundamentals of
Econometrics

Week-14

Week-13:
Previous Lectures on Autocorrelation

1
 1/1/2023

What Next?

Last Weeks

Few Important Things!!!

▪ Few more things that I think is more important to cover before moving to the
content of the Last topics.

▪ Last weeks topics are an overviews of the two very big areas of the
Econometrics

▪ I will reduce this overviews to one/two lectures only given that it cannot be
covered fully in either case

▪ And will cover few more important things that are more important than the
overview you may have in last two weeks.

2
 1/1/2023

CHAPTER 7
Regression Diagnostic IV: Model Specification Error

Introduction to Specification Errors

3
 1/1/2023

Introduction: Model Specification Errors

CLRM assumes that model is correctly specified. By correct specification
we mean one or more of the following
1. The model does not exclude any “core” variables (Under Fitting a Model) Tests
2. The model does not include superfluous variables (Over Fitting a Model) Tests
3. The functional form of the model is suitably chosen
4. There are no errors of measurement in the regressand and regressors
5. Outliers in the data, if any, are taken into account
6. The probability distribution of the error terms is well specified
7. What happens if the regressors are stochastic?
8. The Simultaneous Equation Problem: the simultaneity bias
7

Omission of relevant variables (Under Fitting)

• Suppose True Model is: 𝑌𝑖 = 𝐵1 + 𝐵2 𝑋2𝑖 + 𝐵3 𝑋3𝑖 + 𝑢𝑖
• But for any reason we fit the model as follows; 𝑌𝑖 = 𝐴1 + 𝐴2 𝑋2𝑖 + 𝑣𝑖
• The consequences of omitting variable X3 will be :
i. If X3 is correlated with X2, then estimated value of 𝐴1 and 𝐴2 will be biased
(𝐸(𝑎1 ) ≠ 𝛽1 & 𝐸(𝑎2 ) ≠ 𝛽2 ) and inconsistent , and bias does not disappear as the
sample size gets larger
ii. If X2 and X3 are not correlated ….then 𝑎1 will be biased but 𝑎2 is now unbiased
iii.Variance of error term 𝜎 2 is incorrectly estimated
iv. Variance of 𝑎2 will be a biased estimator of the variance of the true estimator 𝛽2
v. This lead to question the hypothesis testing process and the prediction being made
based on the model

4
 1/1/2023

Inclusion of an Unnecessary Variable(s) (Over Fitting)

• Suppose True Model is: 𝑌𝑖 = 𝐵1 + 𝐵2 𝑋2𝑖 + 𝑢𝑖
• But for any reason we fit the model as follows; 𝑌𝑖 = 𝐴1 + 𝐴2 𝑋2𝑖 + 𝐴3 𝑋3𝑖 + 𝑣𝑖
▪ The consequences of the over Fitting will be as follows;
i. The OLS estimator of the incorrect models are all unbiased and consistent
that is 𝐸(𝑎1 ) = 𝛽1 and 𝐸 𝑎2 = 𝛽2 and E 𝑎3 = 𝛽3 =0
ii. The error variance 𝜎 2 is correctly estimated.
iii. The usual confidence interval and hypothesis-testing procedures remain valid.
iv. However, the estimated 𝑎′ s will be generally inefficient, that is, their
variances will be generally larger than those of the 𝑏 ′ s

Over fitting versus under fitting

Over Fitting of the Model Under Fitting of the Model
▪ Biased
▪ Unbiased
▪ Inconsistent
▪ Consistent
▪ The error variance is incorrectly
▪ The error variance is correctly estimated
estimated
▪ Usual hypothesis-testing procedures
▪ The conventional hypothesis-testing become invalid.
methods are still valid.
Conclusion: The best approach is to
▪ The only penalty for the inclusion of include all relevant explanatory variables
the superfluous variable is that the that, on theoretical grounds, directly
estimated variances of the coefficients influence the dependent variable.
are larger.
Do not draw to go for adding “unnecessary variables” to avoid the problem of relevant
variables …. Being called “kitchen sink approach” by Guajarati.
Such a philosophy is not recommended!!!
10

5
 1/1/2023

Test of Specification Error: Background

▪ Specification Errors may not be deliberately committed
▪ More often these error happens due to:
✓ our inability to formulate the model as precisely as possible
✓ weak underlying theory
✓Unavailability of the right kind of data to test the model

▪ The practical questions is how to detect specification bias

▪ Because if we found that there is a problem of specification
▪ ………we can search for reasons
▪ ………….. and given the reasons …. remedial measures are available

11

Test of Specification Error: Background

➢ We are never sure that the model is true
➢ On the basis of theory and prior empirical work, we develop a model that we
believe captures the essence of the subject under study
➢ Then subject the model to empirical testing
➢ After we obtain the results, we begin the postmortem, keeping in mind the
criteria of a good model
➢ We inspect some broad feature of the model
➢ T-stats, S.E, 𝑅ത 2 , and d statistics
➢ If these diagnostics are reasonably good, we proclaim that the chosen model
is a fair representation of reality

12

6
 1/1/2023

Tests for Omitted Variables

(under fitted Model)
▪ There are several tests of detecting the omission of relevant variables,
but we will consider following two Tests only;
▪ General Test
▪ Ramsey’s RESET test
▪ the Lagrange multiplier (LM) test

13

Test for Model Under Fitting: General Test

This could be the started point for identifying “correct model”
It has following steps:
1. Estimated the original model (restricted) without considering the variables that you
suspect to be included or not to be included. Take its R2. This is R2r
2. Estimate the unrestricted model including 2the ignored variables as an explanatory
variables. And take R2 of this Model. Its R Ur
3. Now we check if “it worth it” to expand the model by adding additional variables.
4. Under the null hypothesis that the restricted (i.e., the original) model is correct, we
can use the F test .
2
(𝑅𝑢𝑟 − 𝑅𝑟2 )ൗ m = # of restriction/ignored variables
𝑚 n = # of observations
𝐹= 2
(1 − 𝑅𝑢𝑟 )൘ k = # of beta in unrestricted model
(𝑛 − 𝑘)
5. If the F test in Step 4 is statistically significant, we can reject the null hypothesis
and it implies that the restricted model is not appropriate in the present situation.
(i.e., we should add the ignored variables in model)

14

7
 1/1/2023

Test for Model Under Fitting: Ramsey’s RESET Test

Ramsey’s regression specification error test or Remsey’s (RESET) is a general test of model
specification errors: It has following steps:
1. From the (incorrectly) estimated model, we first obtain the estimated, or fitted, values of the
dependent variable.
2. Re-estimate the original model including the square and cubic terms of the estimated
dependent variable as an explanatory variables
3. The initial model is the restricted model and the model is Step 2 is the unrestricted model.
4. Under the null hypothesis that the restricted (i.e., the original) model is correct, we can use
the F test . 2
(𝑅𝑢𝑟 − 𝑅𝑟2 )ൗ
𝑚
𝐹= 2 m = # of restriction
(1 − 𝑅𝑢𝑟 )൘ n = # of observations
(𝑛 − 𝑘) k = # of beta in unrestricted model
5. If the F test in Step 4 is statistically significant, we can reject the null hypothesis and it
implies that the restricted model is not appropriate in the present situation. (i.e., we should
add the ignored variables in model)
6. By the same token, if the F statistic is statistically insignificant, we do not reject the original
model.

15

Test for Model Under Fitting: Ramsey’s RESET Test

▪ The RESET test has two drawbacks.
1.If the test shows that the chosen model is incorrectly specified, it does not
suggest any specific alternative.
2.the test does not offer any guidance about the number of powered terms of
the estimated values of the regressand to be included in the unrestricted
model.
▪ There is no definite answer to this, although in practice we could proceed by
trial and error and select the powered terms on the basis of information
criteria, such as AIC/SIC/HQC.

16

8
 1/1/2023

Test for Model Under Fitting: Lagrange Multiplier Test

1. From the original model, we obtain the estimated residuals, ui.
2. If in fact the original model is the correct model, then the residuals ui obtained
from this model should not be related to the regressors omitted from that model.
3. We now regress ui on the regressors in the original model and the omitted variables
from the original model. This is the auxiliary regression.
4. If the sample size is large, it can be shown that n (the sample size) times the R2
obtained from the auxiliary regression follows the chi-square distribution with df
equal to the number of regressors omitted from the original regression.
5. If the computed chi-square value exceeds the critical chi-square value at the chosen
level of significance, or if its p value is sufficiently low, we reject the original (or
restricted) regression.
6. This is to say, that the original model was mis-specified.

17

Example: Detection of over fitted model

➢ Suppose we develop a k-variable model to explain a phenomenon:
𝑌𝑖 = 𝐵1 + 𝐵2 𝑋2𝑖 + 𝐵3 𝑋3𝑖 + ⋯ 𝐵𝑘 𝑋𝑘𝑖 + 𝑢𝑖
➢ But we are not sure if one variable of it (say 𝑋𝑘𝑖 ) really should be in the
model
➢ A simple way to proceed is …. To check the significance of the estimated 𝐵𝑘
with the usual t test
➢ If we are not sure about more than one variable (say, 𝑋3𝑖 and 𝑋4𝑖 ) that if
these should be in the model
➢ This can be checked via F-Test of Remsey’s (RESET)/LM Test and check for
restricted versus unrestricted model to decide if a variable should be added or
not.

18

9
 1/1/2023

Functional form of the model

19

Functional form of the model

➢ We already had discussion on this topic in the CHPATER 2

➢Please check the last part of Lecture 11 (Week6: Lecture 11, LAST 10
minutes had this discussion)

20

10
 1/1/2023

Measurement Errors
➢ Till now we assumed implicitly that the dependent variable Y and the
explanatory variables, the X’s, are measured without any errors.

➢ Furthermore, it not explicitly spelled out in assumption of CLRM, but it is

presumes that the values of the regressand as well as regressors are accurate.
…i.e., they are not guess estimates, extrapolated, interpolated or rounded
off in any systematic manner or recorded with errors

➢ In practice it is not true due to nonresponse errors, reporting errors etc.

➢ Whatever the reasons, error of measurement is a potentially troublesome

problem

21

Errors of Measurement in the (Dependent

Variable/ Regressand)
1. The OLS estimators are still unbiased.
2. The variances and standard errors of OLS estimators are still unbiased.
3. But the estimated variances, and also the standard errors, are larger than
in the absence of such errors.

In short, errors of measurement in the regressand do not pose a very

serious threat to OLS estimation.

22

11
 1/1/2023

Errors of Measurement in the (Independent

Variables/ Regressor)
1. OLS estimators are biased as well as inconsistent.
2. Errors in a single regressor can lead to biased and inconsistent estimates of the
coefficients of the other regressors in the model.
✓ It is not easy to establish the size and direction of bias in the estimated coefficients.
3. It is often suggested that we use instrumental or proxy variables for variables
suspected of having measurement errors.
✓ The proxy variables must satisfy two requirements—that they are highly correlated with
the variables for which they are a proxy and also they are uncorrelated with the usual
equation error as well as the measurement error
✓ But such proxies are not easy to find.
4. We should thus be very careful in collecting the data and making sure that
some obvious errors are eliminated.

23

Outliers, Leverage, And Influence Data

24

12
 1/1/2023

Background
➢ OLS gives equal weight to every observation in the sample.
➢ This may create problems if we have observations that may not be
“typical” of the rest of the sample.
➢ Such observations, or data points, are known as outliers, leverage or
influence points.

25

Outliers, Leverage, and Influence Data

➢ Outliers: In the context of regression analysis, an outlier is an observation with a
large residual (ui), large in comparison with the residuals of the rest of the
observations.
➢ Leverage: An observation is said to exert (high) leverage if it is disproportionately
distant from the bulk of the sample observations. In this case such observation(s)
can pull the regression line towards itself, which may distort the slope of the
regression line.
➢ Influential point: If a levered observation in fact pulls the regression line toward
itself, it is called an influential point. The removal of such a data point(s) from the
sample can dramatically change the slope of the estimated regression line.
▪ A traditional approach is to use error terms and see if it display any far values for
the residuals. One can do analysis by dropping such values for identifying. There
are some tests for that too, that we will exclude at this stage.
26

13
 1/1/2023

The Simultaneity Problem

32

The Simultaneity Problem

➢ There are many situations where unidirectional relationship between Y and the Xs
cannot be maintained, since some Xs affect Y but in turn Y also affects one or more
Xs.
✓ In other words, there may be a feedback relationship between the Y and X variables.
➢ Simultaneous equation regression models are models that take into account feedback
relationships among variables.
➢ Endogenous variables are variables whose values are determined in the model.
➢ Exogenous variables are variables whose values are not determined in the model.
✓ Sometimes, exogenous variables are called predetermined variables, for their values are
determined independently or fixed, such as the tax rates fixed by the government.
➢ Estimate parameters using Method of Indirect Least Squares (ILS) or Method of
Two-Stage Least Squares (2SLS).

33

14
 1/1/2023

Thank You

36

15